1. Field of the Invention
This invention relates to Raman spectrometers characterized by extremely favorable signal to noise ratios and the ability to obtain usable spectral data in situations where Raman spectroscopy has heretofore proved to be unworkable. More particularly, it is concerned with such Raman spectral instrumentation making use of a stationary electrooptical mask in lieu of conventional slit scanning hardware, together with appropriate computer-controlled electronics permitting use of Hadamard multiplexing techniques.
2. Description of the Prior Art
Raman spectroscopy was discovered in 1928 and has been an important method for the elucidation of molecular structure, for locating various functional groups or chemical bonds in molecules, and for the quantitative analysis of complex mixtures, particularly for major components. Although Raman spectra are related to infrared absorption spectra, a Raman spectrum arises in a quite different manner and thus provides complementary information.
When monochromatic light is scattered by molecules, a small fraction of the scattered light is observed to have a different frequency from that of the irradiating light; this is known as the Raman effect. Raman spectroscopy is in turn based upon the Raman effect, and involves the passage of monochromatic light into a sample that contains molecules which can undergo a change in molecular polarizability as they vibrate. It is strictly a quantum effect. Most collisions of the incident photons of the irradiating monochromatic light with the sample molecules are elastic, resulting in so-called Rayleigh scattering. In such Rayleigh scattering, the electric field produced by the polarized molecule oscillates at the same frequency as the incident electromagnetic wave, so that the molecule acts as a source sending out radiation of that frequency. The incident radiation does not raise the molecule to any particular quantized level, and accordingly the molecule can be considered as in a virtual excited state. As the electromagnetic wave passes, the polarized molecule ceases to oscillate and returns to its original ground level in a very short time (approximately 10.sup.-12 seconds).
A small proportion of the excited molecules of the sample (10.sup.-6 or less) may undergo a change in polarizability during one of the normal vibrational modes. This provides the basis for the Raman effect. Usually incident radiation, V.sub.O, is absorbed by a molecule in the lowest vibrational state. If that molecule while in the virtual state, re-emits by returning not to the original vibrational state, but to an excited vibrational level, V.sub.v, of the ground electronic state, the emitted radiation is of lower energy, or lower frequency than the incident radiation. The difference in frequency is equal to a natural vibration frequency of the molecule's ground electronic state. Several such shifted lines (Stoke lines) normally will be observed in a Raman spectrum, corresponding to different vibrations in the molecule. This provides a richly detailed vibrational spectrum of the molecule.
A few of the molecules initially will absorb radiation while they are in an excited vibrational state and will decay to a lower energy level, so that their Raman scattered light will have a higher frequency than the incident radiation. These are called anti-Stokes lines. Thus the spectrum of the scattered light consists of a relatively strong component with frequency unshifted (Rayleight scattering), corresponding to photons scattered without energy exchange, and the two components of the Raman spectrum, namely the Stokes and anti-Stokes lines. Normally, for chemical analysis, only the Stokes lines are considered, because of their greater intensity.
In the usual Raman technique, the excitation frequency of the source radiation is selected to lie below most S-S* electronic transitions and above most fundamental vibration frequencies, although this is not always the case, such as in resonance Raman spectroscopy.
Raman spectroscopy offers several advantages over conventional IR absorption measurements. First, Raman spectroscopy can be used to detect and analyze molecules with infrared inactive spectra, such as homonuclear diatomic molecules. For complicated molecules whose low symmetry does not forbid both Raman and infrared activity, certain vibrational modes are inherently stronger in the Raman effect and weaker in, or apparently absent from, the infrared spectrum. Raman activity tends to be a function of the covalent character of bonds and the molecular polarizability of the molecule. Hence, a Raman spectrum reveals information regarding the structure of the molecule.
Raman spectra can be used to study materials in aqueous solutions, a medium that transmits infrared radiation very poorly. Finally, sample preparation for Raman is normally simpler than for IR absorption.
While Raman spectroscopy does therefore exhibit decided advantages, certain heretofore intractable problems have detracted from the usefulness of the technique. The primary disadvantage is the fluorescent background that is generated upon intense laser radiation of many samples. Relative to the Raman signal, the fluorescent background can be enormous, completely obliterating the spectrum. Even if the Raman spectrum could be observed superimposed on the fluorescent background, the noise contribution of the fluorescent emission degrades the signal to noise ratio of the Raman spectrum.
Most conventional Raman instruments make use of an expensive photomultiplier tube (PMT) as a detector. The PMT is normally required because of the very weak Raman signal, and prior attempts at using inexpensive detectors such as room temperature silicon diodes in conventional dispersive instruments have resulted in unacceptable signal to noise ratios, again to the point of obliterating the spectrum. Thus, the cost of Raman instrumentation can be considerable.
It has previously been demonstrated that the use of a red (Kr.sup.+) or near-infrared (Nd:YAG) exciting laser in Raman spectrometry avoids the problems of sample fluorescence and sample photodecomposition common to blue-green (Ar.sup.+) or higher energy laser Raman spectrometry. However, use of these relative low-energy lasers results in a decrease in the intensity of the Raman scattered radiation. Thus, it was found necessary to employ a Fourier transform multiplex spectrometer in an attempt to recover the loss in the signal to noise ratio due to the weaker Raman scattered radiation. When a Fourier transform spectrometer is used for Raman spectroscopy, however, the multiplex nature of the Fourier transform instrument leads to several problems stemming from the relatively weak intensity of the Raman scattered radiation, and the comparatively strong intensity of the Rayleigh scattered radiation. As a consequence, special efforts have been made in such systems to remove the Rayleight radiation. Such prior efforts involved the use of optical pass filters to remove the Rayleigh line before the radiation was admitted into the instrument. The use of optical pass filters, however, has the disadvantage of significantly decreasing the frequency range of the Raman scattered radiation that may be observed. Moreover, the overall intensity of the Raman scattered radiation is lowered as well.
As noted above, use of Fourier transform spectroscopy in Raman spectral analysis is hampered by the large disparity between the intensities of the Raman radiation and the Rayleigh radiation. These two characteristics of the Raman technique translate into at least two considerations which must be treated in the design of any Raman multiplex instrument (i.e., one where the simultaneous measurement of more than one resolution element of radiation at a time where a resolution element is defines as a short wavenumber interval of radiation).
One consequence of the comparatively strong Rayleigh radiation is that when a multiplex method of data acquisition is used, any noise associated with the Rayleigh line will be distributed throughout the entire spectrum. Since the intensity of the Rayleigh radiation is at least three orders of magnitude greater than the intensity of the Raman radiation, the signal to noise ratio for the entire spectrum will be significantly degraded. Thus, if the Rayleigh line is not removed, the weaker Raman scattering may be partially obscured or completely obliterated.
A second consideration in the design of a multiplexed Raman instrument is the larger dynamic range of the signal on the detection system (a factory of approximately 10.sup.3) due to the difference between the Rayleigh scattered radiation intensity and the Raman scattered radiation intensity. Even assuming that the Rayleigh scattered radiation is noise-free, the detection system of the multiplexing instrument must be capable of discriminating between small changes in the multiplex coding of the radiation intensity.
Thus, while the ability of multiplexing techniques to improve the signal to noise ratio of conventional dispersive spectrometers is understood, conventional multiplex Fourier transform spectrometers present serious difficulties when used in the context of Raman spectroscopy, because of the inability to properly eliminate the Rayleigh line from the spectrum without a concomitant serious degradation of the resultant analytical results.
Another multiplexing technique heretofore developed is known as Hadamard transform spectrometry. The theory of Hadamard data encodement and its application to the field of optical spectrometry has been described by Harwit, et al., Hadamard Transform Optics, Academic Press, New York, 1979. Hadamard transform spectrometers are multi-slit coding devices that select unique combinations of resolution elements of radiation via an encoding mask placed in the exit focal plane of a conventional dispersive spectrometer. The radiation allowed to pass through the encoding mask is collected, dedispersed, and detected with conventional detectors. The selection of resolution elements allowed to pass through the mask is governed by the weighing designs given by Hadamard metrices. It is through the encoding mask that Hadamard transform spectrometry derives its multiplexing capabilities. Typically, a Hadamard mask is computer-controlled, with the appropriate Hadamard mathematics being applied as software in the computer.
Typical encodement masks forming a part of Hadamard devices are multiple-slit, mechanically movable bodies which are periodically exchanged or shifted during a spectral analysis. As a consequence of this design, the resulting spectra are inevitably degraded, inasmuch as it is virtually impossible to control the movement and positioning of mechanical mask with the degree of accuracy commensurate with that of the remainder of the instrument. In short, conventional mechanical masks are very expensive and difficult to properly operate.
Further details pertaining to Raman spectroscopy, multiplexing techniques and Hadamard spectroscopy can be found in the aforementioned Harwit, et al. publication and in the following: Instrumental Methods of Analysis, H. H. Willard, et al., Chapter 8, pp. 217-238; and in U.S. Pat. No. 4,615,619 and the references cited therein. These publications and U.S. Pat. No. 4,615,619 are incorporated by reference herein. The aforementioned patent specifically discloses an improved electrooptical masking device designed to be positioned in a stationary fashion in a dispersive spectrometer to facilitate Hadamarad encodement techniques.